Engineering Systems Design Analytics and AI, 6.0 ECTS
September 18 @ 13:00 - October 30 @ 17:00
CREDITS |
6.0 (ECTS) credits |
EXAMINER |
Petter Krus, Linköping University (LiU) |
CONTACT |
Petter Krus, Linköping University (LiU) petter.krus@liu.se |
TARGET GROUP |
PhD student interested in model-based system engineering and design analytics and the application of artificial intelligence. |
PREREQUISITES |
Basic background in engineering. Some experience in Python is advantageous but not required. |
AIM |
The course aims to equip students with analytical tools for evaluating complex systems, focusing on statistical modelling, machine learning, sensitivity analysis, and computational complexity. It should prepare the participants to take on the transformation implied by the rise and democratization of machine learning and large language models and the ability to use these methods in industrial projects. |
LEARNING OUTCOMES |
After the course, the student shall demonstrate skill and ability in: – Generate models based on statistical data regarding relations between different requirements and system/component characteristics. – Produce sensitivity analysis of a system and be able to draw conclusions regarding robustness and the degree of coupling. – Estimate system complexity and make analysis of computational complexity. – Be able to generate extensive programming using Generative AI. |
CONTENTS |
Design analytics (8h): – Axiomatic design. – Sensitivity analysis. – Robustness. – Functional correlation. Machine learning and design space generation (8h): – Introduction to Machine Learning in design. – Multiple regression analysis. – Principal component analysis (PCA) using Singular value decomposition (SVD). – Neural networks. – Design space and parametrization. System structures and modelling (4h): – The use of Large Language Models (LLMs) for analysis and modeling and simulation. – Requirement and Functional modeling (UML/SysML). – Dependency structure matrix. – System Concept Generation using LLM. Information entropy in design (4h): – Shannon’s Information Theory. – Information entropy as a measure for system complexity. – Design space quantification. – Axiomatic Design and Complexity. – System Optimization and Information Entropy. |
ORGANISATION |
Lectures, projects, and a one-day examination presentation meeting. The project will be a collaborative project using LLMs to generate analysis codes (tentatively) in Python that reflect the content of the course and apply it to an example related to the individual student’s research projects. For each occasion there will be about 2h lectures and 2h of supervised exercises. There will one half day examination presentation meeting. |
LITERATURE |
Course Compendium + Scientific papers |
EXAMINATION |
The main examination task is a written project report that can be aligned towards the PhD students’ research. The grade is passed/not passed. |
SCHEDULE |
– Sept 18. 13-17 (CET). Design analytics I, Axiomatic design, Machine learning in design – Sept 25. 13-17 (CET). Design analytics II. Sensitivity Analysis, Robustness and functional correlation. – Oct. 2. 13-17 (CET). Machine learning and design space generation I: Subsystem modelling. Regression analysis and Principal Component Analysis. – Oct. 9. 13-17 (CET). Machine learning and design space generation II. Principal Component Analysis and Neural Networks. System structures and modelling. – Oct. 16. 13-17 (CET). Information entropy in Design and Optimization. – Oct 30 13-17 (CET). Project presentation.Schedule is preliminary and may be subject to change. |